Give art a chance¶
from IPython.display import Image, HTML
Image('logo.png')
Project: Fountains.
Authors:
Justus Harris (http://www.justusharris.com/work/#/medsculpart/),
Eric Dolores (http://rubiel1.github.io),
Rene Marcelo (https://renemarcelo.github.io).
Image("Fountains1.jpg")
2016: At that time I had interviews with math consulting companies:
sometimes you are working on a project and you realizes that what the company is doing is tring to cover something illegal...
A little bit discouraged, I joined Chihack looking for experience programming in a project aligned with my moral values.
I meet the artist Justus who worked with sculptures from diabetes data:
Health Data Sculptures simplify each month of vital metrics insulin doses, blood glucose levels, and carbohydrates consumed, into objects with textures and shapes that vary based on changes in health felt sequentially as a timeline. Image('Justus.png', width=300)
His sculptures can help people visualize when something is not ok. So after hearing the process of creation, I proposed I could use python to automatize the process in Blender.
After working for a while he invited me to join in an application for an internet of things project dealing with enviromental data.
Image('competition.png')
In the project we took data from lake michigan using a rasberry pi and we created representations of water pollution.
Rene¶
The first part of the project was done by the artist Rene,
Image("rene.jpg")
Image("expo.jpg")
At the left panel we see water with high levels of turbidity while on the right panel we see water with low levels of turbidity. Turbidity is caused by particles and microorganisms.
We webscrapped turbidity data from lake Michigan from the last 6 months and Rene made a song based on the data.
The data was obtained from https://waterdata.usgs.gov, they provide scrapping tools to get easily the data in different formats.
Image('website.png')
Image('rene.png')
The raspberry pi reads the last 6 months of data, and Rene's algorithm creates music that helps recognize when the turbidity level is beyond safe consuption.The image above is an example of the data collected. What we heard is the music that Rene created.
HTML('<iframe width="100%" height="300" scrolling="no" frameborder="no" allow="autoplay" src="https://w.soundcloud.com/player/?url=https%3A//api.soundcloud.com/tracks/333442792&color=%23ff5500&auto_play=true&hide_related=false&show_comments=true&show_user=true&show_reposts=false&show_teaser=true&visual=true"></iframe>')
I am afraid I dont know more details about his work but you can take a look at his notes https://rene.codes/dot_node_red_saic_harris_art/
Justus¶
Image('friends.jpg')
the master mind that organized the whole process,
interviewed people from different countries,
3d scanned the people,
created artistic effects to the animations,
final editions,
finding sponsors, material, etc.Eric¶
Based on the stories of the people that Justus interviewed I made animations using the 3d Scans.
How to choose the topics?¶
We looked at Flint (lead) and turbidity. At that time I was teaching AP stats, chess and AP calculus. I assigned projects to students and Ying Ding decided to sacrifice her youthful in order to analyze water properties of 200 countries. I took a look at her data and that guide us into water scarcity. I mentioned this in her recommendation letter.
Image("lead.png")
The exhibition looked more or less like:
HTML('<iframe width="560" height="315" src="https://www.youtube.com/embed/95KwM7GkOtw?rel=0&start=33" frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>')
What kind of problems did we face?¶
Low difficulty problem:¶
2000 files with the wrong name (solved with a quick python script).
We asked advice in the SAIC and they recommended us to change from Blender to Maya (we did).
We needed to convince people to donate a 3-d scan of their bodies (we promised not to release the files online).
Working cultures... In art, things seems to be not precise, everything looks flexible (like in probability). I admire artist because it seems such a different way of thinking.
Intermediate difficulty:¶
We got a virus just before the exhibition, losing the last 8 hrs of work.
The solution was to use Maya Graph editor to make those editions (syncronization, rythm, etc). This is the part that required a good abstract mind.Image('graph.gif')
Any of those graphs could represent water's density, bobbing speed, angle of the force applied to the object, transparency of the water, etc.
High difficulty problems.¶
The 3d-scanner doesn't work on black people.
We had the interview, the animation, the data, we just couldn't get the body-face image.Isn't this just an illumination issue?
Suppose that a little girl opens her birthday gift to find out that her new toy doesn't work. But the toy does work with white people. Will you tell the girl what tools does she need to improve the illumination of the scene in order to be recognized as a human by the algorithm?
Image('twitt.jpg', width=300)
On the bright side, if AI starts to kill humans, robots may not see people of color...unless we move.We need to retrain algorithms to remove racial bias.¶
If you are interested in this task, join the Algorithmic Justice League: https://www.ajlunited.org/
In our case, we couldn't add our subject to our exhibition.
Diane Khalo showed an exhibition in NU. It was a room where you can see pictures of women that were killed in Ciudad Juarez and read what did they liked, what where their life goals ... that was more powerful that knowing that 1775 women were killed in the last 25 years.
A 1000 children were abused by Catholic priests...
'1000 victims' doesn't reflect the confusion, pain and suffering of the victims.
We learn numbers by using abstraction and so we separate them from feelings.
We need to make sure our algorithms are not biased, and we need people to re-train the current models.
Working with people of different areas is challenging but it is worthy.¶
Minorities don't see representatives working on technology, those representatives exist, but they hide to make their lifes easier. Once you make it, be the leader you needed when you was growing.¶
For example, women have been present in the hearth of computer science since the beginning, in 1965 navajo women were hired to do circuit assembly.
Image('circuit.jpg')
A possible explanation of their excellent performance is that their weaving involved simmilar geometric patterns.
from IPython.display import Image
Image("navajot.jpg")
The gift that keeps on giving¶
I applied to a postdoc position while in Chihack but I didn't get in, then after winning the competition I applied again and I got accepted.
Currently I am helping find Jenny, a Chinese women that visited Mexico and went missing.
Image('jenny.png')
Any information or help will be welcome https://www.facebook.com/groups/helpfindjenny/
Suppose that you suspects a person that is missing could be in the dark web. There is people in Mexico that spends their time comparing images of their database of missing people with pictures on the dark web.
Perhaps we can scrap those images, then use face recognition to compare them with the database of missing people, and then if there is a match send it to the interpool... If any of you is familiar with the topics, or would like to contribute, feel free to contact me twitter:Don_Rubiel.
If you have knolwedge of machine learning and would like to do applied research, think about our group at FSU.¶
http://mendoza.eng.fsu.edu/join.html This is a picture of our group.
Image("group.png", width=400)
And the picture with style transfer using neural networks...
...from an Aztec paint:
Image("style_transfer_aztec.png", width=400)
...from a draw by 迷路 畫家:
Image('style_transfer_new2.png',width = 400)
...from gothic letters:
Image("style_transfer_gothic.png", width=400)
